ECCV 2020 论文列表

Image-to-image translation is affected by entanglement phenomena, which may occur in case of target data encompassing occlusions such as raindrops, dirt, etc. Our unsupervised model-based learning disentangles scene and occlusions, while benefiting from an adversarial pipeline to regress physical parameters of the occlusion model. The experiments demonstrate our method is able to handle varying types of occlusions and generate highly realistic translations, qualitatively and quantitatively outperforming the state-of-the-art on multiple datasets."

Few-shot learning is an important research problem that tackles one of the greatest challenges of machine learning: learning a new task from a limited amount of labeled data. We propose a model-agnostic method that improves the test-time performance of any few-shot learning models with no additional training, and thus is free from the training-test domain gap. Based on only the few support samples in a meta-test task, our method generates the samples adversarial to the base few-shot classifier's boundaries and fine-tunes its embedding function in the direction that increases the classification margins of the adversarial samples. Consequently, the embedding space becomes denser around the labeled samples which makes the classifier robust to query samples. Experimenting on miniImageNet, CIFAR-FS, and FC100, we demonstrate that our method brings significant performance improvement to three different base methods with various properties, and achieves the state-of-the-art performance in a number of few-shot learning tasks."

Unsupervised image classification is a challenging computer vision task. Deep learning-based algorithms have achieved superb results, where the latest approach adopts unified losses from embedding and class assignment processes. Since these processes inherently have different goals, jointly optimizing them may lead to a suboptimal solution. To address this limitation, we propose a novel two-stage algorithm in which an embedding module for pretraining precedes a refining module that concurrently performs embedding and class assignment. Our model outperforms SOTA when tested with multiple datasets, by substantially high accuracy of 81.0% for the CIFAR-10 dataset (i.e., increased by 19.3 percent points), 35.3% accuracy for CIFAR-100-20 (9.6 pp) and 66.5% accuracy for STL-10 (6.9 pp) in unsupervised tasks."

We show that simple patch-based models, such as epitomes (Jojic et al., 2003), can have superior performance to the current state of the art in semantic segmentation and label super-resolution, which uses deep convolutional neural networks. We derive a new training algorithm for epitomes which allows, for the first time, learning from very large data sets and derive a label super-resolution algorithm as a statistical inference algorithm over epitomic representations. We illustrate our methods on land cover mapping and medical image analysis tasks."

Recent studies have shown that, context aggregating information from proposals in different frames can clearly enhance the performance of video object detection. However, these approaches mainly exploit the intra-proposal relation within single video, while ignoring the intra-proposal relation among different videos, which can provide important discriminative cues for recognizing confusing objects. To address the limitation, we propose a novel Inter-Video Proposal Relation module. Based on a concise multi-level triplet selection scheme, this module can learn effective object representations via modeling relations of hard proposals among different videos. Moreover, we design a Hierarchical Video Relation Network (HVR-Net), by integrating intra-video and inter-video proposal relations in a hierarchical fashion. This design can progressively exploit both intra and inter contexts to boost video object detection. We examine our method on the large-scale video object detection benchmark, i.e., ImageNet VID, where HVR-Net achieves the SOTA results. Codes and models will be released afterwards."

This paper studies the problem of learning semantic segmentation from image-level supervision only. Current popular solutions leverage object localization maps from classifiers as supervision signals, and struggle to make the localization maps capture more complete object content. Rather than previous efforts that primarily focus on intra-image information, we address the value of cross-image semantic relations for comprehensive object pattern mining. To achieve this, two neural co-attentions are incorporated into the classifier to complimentarily capture cross-image semantic similarities and differences. In particular, given a pair of training images, one co-attention enforces the classifier to recognize the common semantics from co-attentive objects, while the other one, called contrastive co-attention, drives the classifier to identify the unshared semantics from the rest, uncommon objects. This helps the classifier discover more object patterns and better ground semantics in image regions. In addition to boosting object pattern learning, the co-attention can leverage context from other related images to improve localization map inference, hence eventually benefiting semantic segmentation learning. More essentially, our algorithm provides a unified framework that handles well different WSSS settings, i.e., learning WSSS with (1) precise image-level supervision only, (2) extra simple single-label data, and (3) extra noisy web data. It sets new state-of-the-arts on all these settings, demonstrating well its efficacy and generalizability."

Domain Adaptation (DA) transfers a learning model from a labeled source domain to an unlabeled target domain which follows different distributions. There are a variety of DA scenarios subject to label sets and domain configurations, including closed-set and partial-set DA, as well as multi-source and multi-target DA. It is notable that existing DA methods are generally designed only for a specific scenario, and may underperform for scenarios they are not tailored to. In this paper, we propose Versatile Domain Adaptation (VDA), where one method can handle several different scenarios at the same time. Towards handling it, a more general inductive bias other than the domain alignment should be explored. We delve into a missing piece of existing methods: class confusion, the tendency that a classifier confuses the predictions between the correct and ambiguous classes for target examples, which exists in different scenarios. We unveil that reducing such pair-wise class confusion brings about significant transfer gains. Based on this, we propose a general loss function: Minimum Class Confusion (MCC). It can be characterized by (1) a non-adversarial DA method without explicitly deploying domain alignment, enjoying fast convergence speed; (2) a versatile approach that can handle the four existing scenarios: Closed-Set, Partial-Set, Multi-Source, and Multi-Target DA, outperforming the state-of-the-art methods in these scenarios, especially on the largest and hardest dataset to date (7.3% on DomainNet). Strong performance in the two scenarios proposed in this paper: Multi-Source Partial and Multi-Target Partial DA, further proves its versatility. In addition, it can also be used as a general regularizer that is orthogonal and complementary to a variety of existing DA methods, accelerating convergence and pushing those readily competitive methods to a stronger level. Code is released in https://github.com/thuml/MCC."

The internal geometry of most modern consumer cameras is not adequately described by the perspective projection. Almost all cameras exhibit some radial lens distortion and are equipped with electronic rolling shutter that induces distortions when the camera moves during the image capture. When focal length has not been calibrated offline, the parameters that describe the radial and rolling shutter distortions are usually unknown. While for global shutter cameras, minimal solvers for the absolute camera pose and unknown focal length and radial distortion are available, solvers for the rolling shutter were missing. We present the first minimal solutions for the absolute pose of a rolling shutter camera with unknown rolling shutter parameters, focal length, and radial distortion. Our new minimal solvers combine iterative schemes designed for calibrated rolling shutter cameras with fast generalized eigenvalue and Groebner basis solvers. In a series of experiments, with both synthetic and real data, we show that our new solvers provide accurate estimates of the camera pose, rolling shutter parameters, focal length, and radial distortion parameters."

Adversarial domain adaptation has made tremendous success by learning domain-invariant feature representations. However, conventional adversarial training pushes two domains together and brings uncertainty to feature learning, which deteriorates the discriminability in the target domain. In this paper, we tackle this problem by designing a simple yet effective scheme, namely Asymmetric Adversarial Domain Adaptation (AADA). We notice that source features preserve great feature discriminability due to full supervision, and therefore a novel asymmetric training scheme is designed to keep the source features fixed and encourage the target features approaching to the source features, which best preserves the feature discriminability learned from source labeled data. This is achieved by an autoencoder-based domain discriminator that only embeds the source domain, while the feature extractor learns to deceive the autoencoder by embedding the target domain. Theoretical justifications corroborate that our method minimizes the domain discrepancy and spectral analysis is employed to quantize the improved feature discriminability. Extensive experiments on several benchmarks validate that our method outperforms existing adversarial domain adaptation methods significantly and demonstrates robustness with respect to hyper-parameter sensitivity."

Modern object detection methods can be divided into one-stage approaches and two-stage ones. One-stage detectors are more efficient owing to straightforward architectures, but the two-stage detectors still take the lead in accuracy. Although recent work try to improve the one-stage detectors by imitating the structural design of the two-stage ones, the accuracy gap is still significant. In this paper, we propose MimicDet, a novel and efficient framework to train a one-stage detector by directly mimic the two-stage features, aiming to bridge the accuracy gap between one-stage and two-stage detectors. Unlike conventional mimic methods, MimicDet has a shared backbone for one-stage and two-stage detectors, then it branches into two heads which are well designed to have compatible features for mimicking. Thus MimicDet can be end-to-end trained without the pre-train of the teacher network. And the cost of memory does not increase much, which makes it practical to adopt large networks as backbones. We also make several specialized designs such as dual-path mimicking and staggered feature pyramid to facilitate the mimicking process. Experiments on the challenging COCO detection benchmark demonstrate the effectiveness of MimicDet. It achieves 46.1 mAP with ResNeXt-101 backbone on the COCO test-dev set, which significantly surpasses current state-of-the-art methods."

Microscopy is a powerful visualization tool in biology, enabling the study of cells, tissues, and the fundamental biological processes; yet, the observed images typically suffer from blur and background noise. In this work, we propose a unifying framework of algorithms for Gaussian image deblurring and denoising. These algorithms are based on deep learning techniques for the design of learnable regularizers integrated into the Wiener-Kolmogorov filter. Our extensive experimentation line showcases that the proposed approach achieves a superior quality of image reconstruction and surpasses the solutions that rely either on deep learning or on optimization schemes alone. Augmented with the variance stabilizing transformation, the proposed reconstruction pipeline can also be successfully applied to the problem of Poisson image deblurring, surpassing the state-of-the-art methods. Moreover, several variants of the proposed framework demonstrate competitive performance at low computational complexity, which is of high importance for real-time imaging applications."

Knowledge Distillation (KD) has been one of the most popular used methods to learn a compact model. However, it still suffers from high demand in time and computational resources caused by sequential training pipeline. Furthermore, the soft targets from deeper models do not often serve as good cues for the shallower models due to the gap of compatibility. In this work, we consider these two problems at the sametime. Specifically, we propose that better soft targets with higher compatibility can be generated by using a label generator to fuse the featuremaps from deeper stages in a top-down manner, and we can employ the meta-learning technique to optimize this label generator. Utilizing the soft targets learned from the intermediate feature maps of the model, we can achieve better self-boosting of the network in comparison with the state-of-the-art. The experiments are conducted on two standard classi-fication benchmarks, namely CIFAR-100 and ILSVRC2012. We test various network architectures to show the generalizability of our MetaDistiller. The experiments results on two datasets strongly demonstrate theeffectiveness of our method."

Remote heart rate estimation is the measurement of heart rate without any physical contact with the subject and is accomplished using remote photoplethysmography (rPPG) in this work. rPPG signals are usually collected using a video camera with a limitation of being sensitive to multiple contributing factors, e.g. variation in skin tone, lighting condition and facial structure. End-to-end supervised learning approach performs well when training data is abundant, covering a distribution that doesn't deviate too much from the distribution of testing data or during deployment. To cope with the unforeseeable distributional changes during deployment, we propose a transductive meta-learner that takes unlabeled samples during testing (deployment) for a self-supervised weight adjustment (also known as transductive inference), providing fast adaptation to the distributional changes. Using this approach, we achieve state-of-the-art performance on MAHNOB-HCI and UBFC-rPPG."

Generation of synthetic data has allowed Machine Learning practitioners to bypass the need for costly collection and labeling of large datasets. Unfortunately the generation of such data often requires experts to carefully design sampling procedures that guarantee creation of realistic scenes. These sampling procedures typically require experts to specify certain structural aspects such as scene layout information. This is often hard to design, as scenes are generally highly complex and diverse. In this paper, we propose a generative model of synthetic scenes that reduces the distribution gap between the scene structure of generated scenes and a real target image dataset. Importantly, since labeling scene structures for real images is incredibly cumbersome, our method operates without any ground truth structure information for real data. Experiments on two synthetic datasets and a real driving dataset show that the method successfully bridges the distribution gap of discrete structural features between real and generated images."

Meta-learning is a powerful paradigm for few-shot learning. Although with remarkable success witnessed in many applications, the existing optimization based meta-learning models with over-parameterized neural networks have been evidenced to ovetfit on training tasks. To remedy this deficiency, we propose a network pruning based meta-learning approach for overfitting reduction via explicitly controlling the capacity of network. A uniform concentration analysis reveals the benefit of capacity constraint for reducing generalization gap of the proposed meta-learner. We have implemented our approach on top of Reptile assembled with two network pruning routines: Dense-Sparse-Dense (DSD) and Iterative Hard Thresholding (IHT). Extensive experimental results on benchmark datasets with different over-parameterized deep networks demonstrate that our method not only effectively alleviates meta-overfitting but also in many cases improves the overall generalization performance when applied to few-shot classification tasks."

We present an interesting and challenging dataset that features a large number of scenes with messy tables captured from multiple camera views. Each scene in this dataset is highly complex, containing multiple object instances that could be identical, stacked and occluded by other instances. The key challenge is to associate all instances given the RGB image of all views. The seemingly simple task surprisingly fails many popular methods or heuristics that we assume good performance in object association. The dataset challenges existing methods in mining subtle appearance differences, reasoning based on contexts, and fusing appearance with geometric cues for establishing an association. We report interesting findings with some popular baselines, and discuss how this dataset could help inspire new problems and catalyse more robust formulations to tackle real-world instance association problems.ootnote{project page: extcolor{magenta}{\url{https://caizhongang.github.io/projects/MessyTable/}}.}"

We are interested in reconstructing the mesh representation of object surfaces from a point cloud. Surface reconstruction is a prerequisite for down-stream applications such as rendering, collision avoidance for planning, animation, etc. However, the task is challenging if the input point cloud has a low resolution, which is common in real-world scenarios (e.g., from LiDAR or Kinect sensors). Existing learning-based mesh generative methods mostly predict the surface by first building a shape embedding that is at the whole object level, a design that causes issues in generating fine-grained details and generalizing to unseen categories. Instead, we propose to leverage the input point cloud as much as possible, by only adding connectivity information to existing points. Particularly, we predict which triplets of points should form faces. Our key innovation is a surrogate of local connectivity, calculated by comparing the intrinsic/extrinsic metrics. We learn to predict this surrogate using a deep point cloud network and then feed it to an efficient post-processing module for high-quality mesh generation. We demonstrate that our method can not only preserve details, handle ambiguous structures, but also possess strong generalizability to unseen categories by experiments on synthetic and real data. "

The objective of this paper is self-supervised learning from video, in particular for representations for action recognition. We make the following contributions: (i) We propose a new architecture and learning framework Memory-augmented Dense Predictive Coding (MemDPC) for the task. It is trained with a predictive attention mechanism over the set of compressed memories, such that any future states can always be constructed by a convex combination of the condense representations, allowing to make multiple hypotheses efficiently.(ii) We investigate visual-only self-supervised video representation learning from RGB frames, or from unsupervised optical flow, or both. (iii) We thoroughly evaluate the quality of learnt representation on four different downstream tasks: action recognition, video retrieval, learning with scarce annotations, and unintentional action classification. In all cases, we demonstrate state-of-the-art or comparable performance over other approaches with orders of magnitude fewer training data."

We introduce an approach for incremental learning that preserves feature descriptors of training images from previously learned classes, instead of the images themselves, unlike most existing work. Keeping the much lower-dimensional feature embeddings of images reduces the memory footprint significantly. We assume that the model is updated incrementally for new classes as new data becomes available sequentially.This requires adapting the previously stored feature vectors to the updated feature space without having access to the corresponding original training images. Feature adaptation is learned with a multi-layer perceptron, which is trained on feature pairs corresponding to the outputs of the original and updated network on a training image. We validate experimentally that such a transformation generalizes well to the features of the previous set of classes, and maps features to a discriminative subspace in the feature space. As a result, the classifier is optimized jointly over new and old classes without requiring old class images. Experimental results show that our method achieves state-of-the-art classification accuracy in incremental learning benchmarks, while having at least an order of magnitude lower memory footprint compared to image-preserving strategies."